Spatiotemporal {Scene} {Interpretation} of {Space} {Videos} via {Deep} {Neural} {Network} and {Tracklet} {Analysis}